Statistically, Is The MLS Season Engine Purring On All Cylinders?

Passes like laser beams…

Alex Olshansky with a look at the quality of play in MLS. Is the league in midseason form?

Olshansky also writes on his own blog at Tempo Free Soccer.

They played well ….

poorly despite …. because they are or are not in mid-season form.

We hear a variation of that quite frequently, especially in the context of the European summer barnstorming tours or CONCACAF Champions League matches. Likewise, soccer blog comment sections routinely fill up with How could he exclude/include him? He is/is not in mid-season form! when discussing a national team’s call-ups.

But what is “mid-season form?” Does it exist? Can we quantify it? Intuitively, the concept rings true. Teams and players benefit from playing every day alongside each other and consistently in games against tough opposition. Decisions are made quicker, passes are released faster, and technical skills are tightened up. It is certainly an environment more conducive to being at peak ability than, say, drinking a Bordeaux and eating oysters on a yacht by Cap d’Antibes (non-MLS players at least).

If this season in MLS is any indication then “mid-season form” certainly does exist. In fact, it seems MLS fans are being treated to an ever-improving level of play as the campaign has progressed.

Before looking at the charts below, it is important to explain the methodology behind what is deemed a “turnover” and a “possession.” The starting point is the OPTA MLS Chalkboards and the “Tackled, Possession Lost” (TPL) metric. TPL is assessed for any errant pass, interception, failed dribble, etc. 

The second component in determining turnovers is the “clearance” metric. All clearances are TPL, but not all clearances are changes in possession. For example, if a defender clears the ball out of bounds, then the team that originally lost the ball never really lost possession. Therefore, the number of clearances, less clearances where the defense maintains possession, is subtracted out of the TPL total. This subtotal is turnovers committed. Attempts on goal are added to turnovers to arrive at a total number of possessions.

First, let us look at the gross data. This is every MLS team’s turnover total for every game of the year.

Exhibit A.


There is a lot going on here. It certainly looks like there is a drop in the number of turnovers around Game 18, but there is a lot of noise in this data set. Let us see if we can make a little more sense of what we are looking at.

All charts below are MLS-wide averages and numbers are on a per team basis.

Exhibit B.

Okay. This confirms what we saw above. Turnovers are down a lot. But that makes sense. It got hot as the spring turned to summer and teams may play a more deliberate style as a result. Well, what about attempts on goal? If possessions are down, they must be down as well.

Exhibit C.


Nope. Attempts on goal have been flat and relatively consistent throughout the entire year. Goals have been pretty flat as well. In fact, goals have actually been slightly higher during the second half of the year than the first. In all, we are seeing about 15-20 less turnovers per team per game (30-40 less total per game) than we were at the beginning of the year! This makes for a much more enjoyable product for neutrals.

Exhibit D.


Exhibit E.

It seems the question is answered.

“Mid-season form” exists. And it is worth at least one less turnover every 2-3 minutes during an average MLS game. It will be interesting to see whether this trend continues as we get to the end of the season or if it plateaus at some point. It would also be worth investigating if this is just an MLS phenomenon, or if this is consistent with other leagues as well.

13 responses to this post.

  1. Posted by Lampard in the End Zone on 2012/08/20 at 5:07 AM

    Good analysis. For the nerds out there, myself included, would you put the R-squared values on the graphs with trendlines?


    • Posted by Alex Olshansky on 2012/08/20 at 9:06 AM

      R^2 of Turnovers Per Attempt was .84. R^2 of Turnovers Per Goal was .30.

      Considered conducting T-tests as well on some of these totals, but find it self-evident that these trends are significant.


      • Posted by Lampard in the End Zone on 2012/08/20 at 9:18 AM

        Thanks! I would hazard a guess that the (non)goodness of fit for Turnovers per Goal might simply might be due to a lack of observations. Perhaps data from a few years would let the Law of Large Numbers kick in.

        As for the T-tests, i readily agree. No need to crunch the numbers.


  2. Is there anyway you could put this spread sheet on google docs so some of us could analyze each team individually?


    • Would love to do that… eventually. All of my data is pulled from my MLS table which I update at my blog.

      Unfortunately, everything is linked in a labyrinth of excel sheets such that to untangle it and put into a google doc isn’t feasible for this season. I’d like to do it in future seasons however.


  3. Posted by tysker on 2012/08/20 at 9:40 AM

    Fun stuff.
    Can you carve out Forced TOs versus Un-Forced TOs? It seems as Defenses approach ‘mid-season form’ they could create more turnovers. Would this offset the quality ‘mid-season form’ of the Offense?


    • Posted by Alex Olshansky on 2012/08/20 at 2:31 PM

      Good point.

      It is almost impossible to break out Forced TO’s and Un-Forced TO’s right now with what OPTA tracks. However, the # of clearances (could be construed as Forced TO’s) is up which suggests that the decrease in turnovers is due to a mixture of improved offensive play AND defensive play. Even though the increase is relatively small, for some reason I neglected to include this in the analysis.


  4. Posted by boohowdy on 2012/08/20 at 10:09 AM

    Great post.


  5. Posted by SurfNdaVe on 2012/08/20 at 10:11 AM

    Oh, the possibilities of data analysis! Love It!


  6. Posted by CaliforniaRedskins on 2012/08/20 at 1:59 PM

    Nice look into this hypothesis. It definitely confirms what you’re seeing as a whole as the season progresses.


  7. Posted by paul on 2012/08/20 at 6:18 PM

    WE NEED MORE OF THIS IN THE US SOCCER BLOGOSPHERE! (Yes, dammit, the caps are necessary.)


  8. Very interesting stuff, Alex. The data opened up by MCFC Analysis should help us check for the trend in the EPL sooner rather than later. For those that don’t know, the head of Man City’s analysis team, Gavin Fleig, has made the 2011-12 EPL Opta stats available to anyone that asks (and promises distribution rights of their work to MCFC and Opta).
    We’re all a bit stuck guessing at cause and effect of this, but fascinating insight nonetheless. No clubs seem to be improving markedly more or less than others, right? You said on twitter that Vancouver showed the same trend despite midseason roster upheaval. What about Toronto and Portland, any shift or lack thereof around coaching changes?


    • Well, every team in the MLS has shown a similar shape. HOWEVER, Toronto does seem to have a less pronounced ‘improvement’ than almost any other team. They were the last MLS team to record a sub-100 turnover game (just did so against Sporting Kansas City). Their turnovers are downward sloping, but more slight.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: